The Data

In a previous analysis, I explored publicly available data from the Bureau of Meteorology to examine change in Australian temperatures. Here, we extend upon this data using gganimate to create three dynamic plots

Let’s load our dependencies

require(tidyverse)
require(readr)
require(ggplot2)
require(gganimate)
require(viridis)
require(scales)
require(gifski)
require(png)
require(transformr)
require(kableExtra)

& read-in the pre-cleaned temperature data from the previous kernel

tbl <- read.csv("ACORN-SAT-Clean.csv")
location2 site.name date minimum.temperature..degC. maximum.temperature..degC. Year Month Week Season Yr_Month Yr_Week Era lat long elev State
1019 KALUMBURU 1941-09-01 20.6 29.8 1941 9 36 Spring 1941-09 1941-36 1940s -14.3 126.65 23 WA
1019 KALUMBURU 1941-09-02 20.6 29.8 1941 9 36 Spring 1941-09 1941-36 1940s -14.3 126.65 23 WA
1019 KALUMBURU 1941-09-03 19.6 29.3 1941 9 36 Spring 1941-09 1941-36 1940s -14.3 126.65 23 WA
1019 KALUMBURU 1941-09-04 21.2 37.1 1941 9 36 Spring 1941-09 1941-36 1940s -14.3 126.65 23 WA
1019 KALUMBURU 1941-09-05 19.8 30.9 1941 9 36 Spring 1941-09 1941-36 1940s -14.3 126.65 23 WA
1019 KALUMBURU 1941-09-06 19.8 NA 1941 9 36 Spring 1941-09 1941-36 1940s -14.3 126.65 23 WA

Before beginning, if using a mac, you may want to consult this resource to do some required set-up in Terminal. Otherwise, we should be good-to-go.

Plot [1]

Monthly average-maximum temperatures, animated by year

In one piece of code, lets aggregate the data & create our plot

# Minimalistic theme for visualisation 
theme_plot_text <-
  theme(
    plot.title = element_text(size = 10, hjust = 0.5),
    axis.title = element_text(size = 10),
    legend.title = element_text(size = 10),
    axis.text = element_text(size = 7))

# plot
month.plot <- tbl %>% 
  group_by(Month, Year) %>% 
  filter(Year != 2019) %>% 
  summarise(avgmax = mean(maximum.temperature..degC., na.rm = TRUE)) %>% 
  ggplot(aes(x = Month, 
             y = avgmax,
             group = Year,
             colour = avgmax)) +
  geom_point() +
  geom_line() +
  scale_colour_gradient2(low = "#E1B823", mid = "#DD8033" , high = "#a4111e",
                         midpoint = 25) +
  labs(title = 'Average daily-maximum temperature in Australia over the past 100 years',
       subtitle = 'Separated by month',
       caption  = "Data Source: ACORN-SAT",
       y = 'Temperature (°C)',
       x = "Year") +
  theme_bw() +
  theme(panel.border = element_blank(),
        axis.line = element_line(colour = "black", size = 0.2),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        plot.title = element_text(size = 6, hjust = 0.5),
        legend.text = element_text(size = 6)) +
  theme_plot_text +
  labs(color = '°C', size = 6)

To animate this plot, we will use “transition states” from package gganimate. This function attributes values of a dimension (in this case, year) to an individual frame. We can then tweak parameters of the animation (length of transitions, whether the plot animation loops, etc)

month.plot <- month.plot + 
  transition_states(states = Year,
                    transition_length = 3, # relative length of transistion
                    state_length = 1, # relative length of pause at the states
                    wrap = TRUE) + # loops the animation
labs(subtitle  = 'Year: {closest_state}') + # includes dynamic sub-title
shadow_mark()

Finally we can animate & view the plot, setting parameters such as framers per second, total frames in the animation, & number of frames comprising the end pause

Following this formula, lets create two further plots

I have taken inspiration from an approach to animated time-series plots by one of R’s gurus Thomas Lin Pedersen

The first of these plots will look at temperature change by month of the year, whilst the second plot will examine by region of Australia

Plot [2]

Time-series of average max temperature, animated by month

Let’s prep the data for plotting

# Create table
temp.max3 <- tbl %>% 
  group_by(Month, Era) %>% 
  filter(Era != 2019) %>% 
  summarise(avgmax = mean(maximum.temperature..degC., na.rm = TRUE))

# Convert numeric months to text 
temp.max3[temp.max3=="12"] <- "December"
temp.max3[temp.max3=="1"] <- "January"
temp.max3[temp.max3=="2"] <- "February"
temp.max3[temp.max3=="3"] <- "March"
temp.max3[temp.max3=="4"] <- "April"
temp.max3[temp.max3=="5"] <- "May"
temp.max3[temp.max3=="6"] <- "June"
temp.max3[temp.max3=="7"] <- "July"
temp.max3[temp.max3=="8"] <- "August"
temp.max3[temp.max3=="9"] <- "September"
temp.max3[temp.max3=="10"] <- "October"
temp.max3[temp.max3=="11"] <- "November"

# as character
temp.max3$Era <- as.character(temp.max3$Era)
# remove s
temp.max3$Era = substr(temp.max3$Era,1,nchar(temp.max3$Era)-1)
# to numeric
temp.max3$Era <- as.numeric(temp.max3$Era)

& create a custom colour palette for the plot

group.colors.gradient <- c("January" = "#ab1312",
                           "February" = "#b01e13", 
                           "March" = "#b83616", 
                           "April" = "#ca6c1b", 
                           "May" = "#d4891e", 
                           "June" = "#dda821", 
                           "July" = "#e1b823", 
                           "August" = "#d89820",  
                           "September" = "#cf7a1d", 
                           "October" = "#c65e1a", 
                           "November" = "#bd4317", 
                           "December" = "#b42a14")

Create our month plot

month.plot <- 
  ggplot(temp.max3, aes(Era, avgmax, group = Month, color = Month)) + 
  geom_line() + 
  geom_segment(aes(xend = 2018, yend = avgmax), linetype = 2, colour = 'grey') + 
  geom_point(size = 2) + 
  geom_text(aes(x = 2018, label = Month), hjust = 0) + 
  scale_colour_manual(values = group.colors.gradient) +
  transition_reveal(Era) + 
  coord_cartesian(clip = 'off') + 
  labs(title = 'Average daily-maximum temperature in Australia over the past 100 years',
       subtitle = 'Separated by month',
       caption  = "Data Source: ACORN-SAT",
       y = 'Temperature (°C)',
       x = "Year") +
  theme_minimal() + 
  theme(legend.position = "none") +
  theme(plot.margin = margin(5.5, 40, 5.5, 5.5))

Plot [3]

Time-series of average max temperature, animated by state

Let’s prep the data for plotting

# Create table
temp.max3C <- tbl %>% 
  group_by(State, Era) %>% 
  filter(Era != 2019) %>% 
  summarise(avgmax = mean(maximum.temperature..degC., na.rm = TRUE))

# as character
temp.max3C$Era <- as.character(temp.max3C$Era)
# remove s
temp.max3C$Era = substr(temp.max3C$Era,1,nchar(temp.max3C$Era)-1)
# to numeric
temp.max3C$Era <- as.numeric(temp.max3C$Era)

& create a custom colour palette for the plot

state.colors.gradient <- 
  c("NT" = "#ab1312", 
    "TAS" = "#e1b823", 
    "VIC" = "#d89820",  
    "SA" = "#cf7a1d", 
    "NSW" = "#c65e1a", 
    "WA" = "#bd4317", 
    "QLD" = "#b42a14")

Create our state plot

state.plot <- 
  ggplot(temp.max3C, aes(Era, avgmax, group = State, color = State)) + 
  geom_line() + 
  geom_segment(aes(xend = 2018, yend = avgmax), linetype = 2, colour = 'grey') + 
  geom_point(size = 2) + 
  geom_text(aes(x = 2018, label = State), hjust = 0) + 
  scale_colour_manual(values = state.colors.gradient) +
  coord_cartesian(clip = 'off') + 
  labs(title = 'Average daily-maximum temperature in Australia over the past 100 years',
       subtitle = 'Separated by State',
       caption  = "Data Source: ACORN-SAT",
       y = 'Temperature (°C)',
       X = "Year") +
  theme_minimal() + 
  theme(legend.position = "none") +
  theme(plot.margin = margin(5.5, 40, 5.5, 5.5)) +
  transition_reveal(Era)